[ad_1]
in Fast Company, Gary Smith and Jeffrey Funk have poured cold water on claims that artificial intelligence will fundamentally revolutionize science.
There wonderful The hype is rampant. Smith and Funk Wired In 2008, he argued that “the deluge of data is making the scientific method obsolete.”
The everyday reality is this: AI will speed up many tedious processes, but it will not replace creative thinking.
Today, AI is increasingly integrated into scientific discovery, accelerating research and helping scientists develop hypotheses, design experiments, collect and interpret large datasets, and write papers. But the reality is that science and AI have very little in common, and AI is unlikely to make science obsolete. At the heart of science are theoretical models that anyone can use to make reliable explanations and predictions.
Gary Smith and Jeffrey Funk, “Why AI Can’t Replace Science,” Fast Company, June 28, 2024
But of course, hype moves faster than progress: For example, former Google CEO Eric Schmidt predicted that large-scale language models (LLMs, or chatbots) could write summaries of papers for scientists. Well, in a way, they were right.
By now we know that LLM literature reviews cannot be trusted. In May 2023, two months before Schmidt’s article was published, a gullible lawyer filed an oral brief written primarily by ChatGPT in a Manhattan court. When ChatGPT was pressed about the fake citations it included in its filing, ChatGPT responded by creating fake details of a fake case. The judge, familiar with the relevant case law, reprimanded (and later fined) the lawyer for filing an oral brief full of “fake judicial decisions…fake citations and fake internal citations.” That, in a nutshell, is the problem with relying on an LLM for literature reviews and other factual information. If you know the facts, you don’t need an LLM. If you don’t know the facts, you can’t trust an LLM.
Smith, Funk, “Nothing can replace science”
This is similar to the recent space bear problem, in which a chatbot generated a lot of utter nonsense about the Soviet Union sending bears into space. As Smith and Funk say, “LLMs tend to generate self-confident garbage.” Since they’re not actually thinking, it’s not clear that LLMs will be any better.
They go on to describe other examples of hype (“AI changes everything”) and slow reality in drug discovery, noting that “the real test will be whether AI enables new products and services to be developed faster and cheaper than without it.” That part remains to be seen, because only a small percentage of problems in science are made easier by more powerful numerical computations.
One thing is certain: hype has much more practical value in show business than it does in science.
Read also: AI is still a delusion. Following instructions and performing fast, tireless, error-free calculations is not intelligence in any sense of the word. Neither OpenAI’s ChatGPT 3.5, Microsoft’s Copilot, nor Google’s Gemini can run simple logic tests because they don’t understand the meaning of the word.
[ad_2]
Source link
